Add quantized torch mean implementation (#27675)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/27675
This leverages QNNPACK global average pooling to perform torch.mean on input feature maps
Currently can only support mean along HxW plane in NCHW tensor.
Test Plan:
python test/test_quantized.py TestQuantizedOps.test_mean
Imported from OSS
Differential Revision: D17989336
fbshipit-source-id: 8d4cbcbed5f146290b1580d26e5b45359d293761